simtk.org

Hello,

I have a number of SO runs that originate from lab results from both a male and a female subject.
During the SO runs, I generally see good results during the simulation (constraint violations under 1e-10 and performance numbers in the single digits).
However, I noticed that in both subjects I'm getting large activation levels (they stick to 100% activation) for the gluteus medius, minimus and soleus muscles. This happens during the end of the stance phase of the leg in question (and consequently during the heel contact of the opposite foot).
The strange part here (for me) is the gluteus maximus show little activity.

This is an example of my male gait runs:
And this is an example of my female gait runs:
I know the colors start mixing once there are several muscles in each image, but the muscles topping 100% in the male result are the gluteus medius 1 and soleus (left leg), whereas in the female result the muscles on 100% are all the gluteus medius and minimus, the soleus and the tensor fascia latae (tfl_l) from the left leg.

Another weird fact is that the female weights only 52Kg, whereas the male weights 70Kg. So given the same max isometric force on both models (both are using the gait2392 standard levels) shouldn't there be much less "effort" per muscle in the female model?


Is this a normal occurrence? Could anyone point me to where I could be doing something wrong? The male model was scaled using OpenSIM's scale tool (with markers), but the female model was scaled using the labs own tool called Visual3D that ports the model directly to *.osim


Thank you.

Another weird fact is that the female weights only 52Kg, whereas the male weights 70Kg. So given the same max isometric force on both models (both are using the gait2392 standard levels) shouldn't there be much less "effort" per muscle in the female model?

If the kinematics are identical and the ground reaction forces are proportional to the weight of the subject then the effort should be less. I can thing of two possible causes of your problem (assuming that kinematics are similar): 1) ground reaction forces are larger than expected for the female, 2) something is wrong with the scaled model.

Another note is that, it is not a good practice for a muscle to saturate. You need to investigate the cause of this by adding reserve actuators and observing which reserves are high and understand the nature of the problem.

Hi and thank you for the response.
I searched relentlessly for something out-of-place with the female's scaled model and I can't seem to find anything wrong.
The female's Pelvis is proportionally larger (anatomically, the female's Pelvis is wider, but the lab converter only does x=y=z scaling), but I can't really see how that would make much of a difference because the model's still symmetric and the muscles seem to be in place.
Afterwards I went to see the reaction forces and, while at first sight they don't seem to show any problems (maximum VY is close to the female's weight), the resulting moment for pelvis tilt taken from Inverse Dynamics or RRA and the respective residual MZ are showing large values.
This is the result of running RRA from the gait2392 reference data:
And these are the female's equivalent results:
But these results also change considerably if I change from point forces to body forces:
(note: I locked the mtp and subtalar angles everywhere - model, tracking tasks, actuators, etc. - because the markers only measured each foot as a whole)

So, from here.. I am completely lost. I think I may have found the root of the problem, but I have no idea how to proceed...
CMC won't run either ("optimizer could not find a solution at time = 0.44), probably because of this. I'm not intending to use CMC as the bulk of my simulations because I'm only looking to study at muscle force/activation.
Any small hint would be greatly appreciated.

Can you provide the figures for the following:

1) Reaction forces (Male vs Female)
2) ID results for both

But these results also change considerably if I change from point forces to body forces:

What do you mean?

By the way you don't have to perform RRA. Scaling -> IK -> SO.

Best

Reaction forces (Male vs Female)

Note: Female weighs ~53Kg, Male weighs ~64Kg

ID results for both

All moments or only the ones related to the pelvis?

What do you mean?

The choice between using the point data or not in the external forces configuration window:

By the way you don't have to perform RRA. Scaling -> IK -> SO.

I'm aware of that, but the OpenSIM documentation suggest that RRA is a good match for SO, and I'm getting noisy results (50-100Hz noise in my SO curves) so I thought RRA would help me get smoother results.


Thank you!

Any possible follow up on this?

I understand it's August and there's a very high chance people are on vacation, but I'd really appreciate some help whenever it's convenient, since I'm still stuck on this...

Hi,

Sorry guys but I'm still in a pinch...

Any help, just a hint on where to look at would be really appreciated.

Thanks!

There might be some useful information in TipsForDebuggingMuscleActuatedSimulations.pdf (linked on the "Getting Started with CMC" page in the Confluence documentation, though the tips apply to Static as well; here is a direct link: https://simtk-confluence.stanford.edu:8 ... 704&api=v2). Some things to check if your muscles are weak: maximum isometric forces (too low?), fiber lengths (operating too far from optimal?), passive forces (too high?), moment arms (too small?).

Thank you for your reply Tom!

Just one question: to what model should I compare these parameters (maximum isometric forces, fiver lengths, passive forces, moment arms) with?

Are the gait2392 results a good starting point or should I seek others?

to what model should I compare these parameters (maximum isometric forces, fiver lengths, passive forces, moment arms) with?

You might check this paper and its references:
Rajagopal, A., Dembia, C.L., DeMers, M.S., Delp, D.D., Hicks, J.L., Delp, S.L. Full body musculoskeletal model for muscle-driven simulation of human gait. IEEE Transactions on Biomedical Engineering, Vol. 63 (9), 2016. http://nmbl.stanford.edu/wp-content/upl ... al2016.pdf

This paper may contain some useful information as well:
Hicks, J.L., Uchida, T.K., Seth, A., Rajagopal, A., Delp, S.L. Is my model good enough? Best practices for verification and validation of musculoskeletal models and simulations of movement. ASME Journal of Biomechanical Engineering 137(2):020905, 2015. http://nmbl.stanford.edu/publications/pdf/Hicks2015.pdf